Introduction

Two of the primary reasons for replicating data between MariaDB servers are to provide greater performance and more redundancy. The traditional master-slave replication covered in Chapter 6, Replication in MariaDB, provides for great read performance by having several read-only slave servers. However, it only solves the redundancy issue partially. In classic replication, there is only one master server node, and if it fails, then one of the slave server nodes must be promoted to become a master server node for the others. Getting this to work correctly in an automated way is difficult.

An easier way to configure replication will be if every node was a master server node. Reads and writes can happen to any of the nodes and the replication component will make sure that everything just works.

MariaDB Galera Cluster makes this sort of replication easy to set up and use. Every node in a Galera Cluster is equal, so if any single node fails it is alright. The cluster will continue running...